SciQuiry SmartSeqQ

SciQuiry is an AI-driven eLearning platform tailored for K-8 science assessments, designed to dynamically adapt to each student's individual learning needs. Utilizing our proprietary AI algorithm, SmartSeqQ, based on Deep Knowledge Tracing (DKT), the platform intelligently adjusts the difficulty and content of questions according to the student's previous answers. This approach not only enhances learning engagement and retention but also ensures that assessments are personalized and growth-focused.

To promote inclusivity, SciQuiry incorporates AI accessibility features such as text-to-speech and speech-to-text technologies, making the platform usable for students with disabilities, including those with visual impairments, dyslexia, and motor impairments. These features help in creating an equitable learning environment for all students.

The content and questions on SciQuiry are generated using generative AI, which speeds up the production of educational materials while maintaining high pedagogical value. All content undergoes rigorous review by subject matter experts before integration to ensure accuracy and relevance.

As SciQuiry operates entirely online, it is accessible via any standard web browser on different devices and screen sizes like desktop computers, tablets, and smartphones. Its compatibility with various devices ensures that it can be easily integrated into school systems or other educational setups with minimal effort. The platform supports both formal education and informal self-guided exploration, fitting seamlessly into existing educational frameworks for in-class and remote learning. 

SciQuiry's core engine, based on neuroscience and educational psychology, leverages active recall and spaced repetition. This method strengthens memory retention and promotes cognitive engagement, requiring students to actively retrieve information to answer questions. The Skill Tree feature utilizes Socratic questioning to progress learners from basic to more complex topics, boosting their problem-solving skills and reinforcing content mastery.

User Case Example: Two middle school students, Student A and Student B, explore the concept of "Distance and Displacement" in Physics through our platform. Both students start with a question that tests their existing knowledge. The platform then customizes the following questions based on their responses, providing a tailored learning experience. Student A quickly grasps the concepts, answering correctly and moving on to more advanced questions. Meanwhile, Student B faces challenges, requiring additional support. The platform adapts, offering detailed explanations for incorrect answers and cool facts for correct answers. Hints are also available for those who struggle. This personalized approach ensures that while Student A progresses faster, Student B receives the guidance needed to build understanding and ultimately grasp the same content as Student A.

Product Demo: https://youtu.be/_j3NVtXc3a8

Our platform is fully accessible via standard web browsers on various devices, ensuring that students can access learning materials from anywhere, removing hardware limitations that often hinder the accessibility of quality educational resources.

The core of SciQuiry's mission is to serve students who are traditionally underserved in educational settings, including those from economically disadvantaged backgrounds and students with disabilities. Our freemium model ensures that essential educational content is available to all students at no cost, providing equitable access to quality education. For those who can afford it, enhanced features are available through a subscription, but no student is ever denied access to the basic learning tools they need to succeed.

We have integrated advanced accessibility features, such as text-to-speech and speech-to-text, into SciQuiry to support students with disabilities, including those with visual impairments and dyslexia. These features are crucial in creating an equitable learning environment where every student has the opportunity to excel.

SciQuiry utilizes AI to tailor the learning experience to the pace and style of each individual student. Our AI algorithms dynamically adjust educational content, mimicking the personalized attention of one-on-one tutoring. This adaptation not only keeps students engaged but also addresses their unique learning needs and strengths, which is particularly critical for learners who might not fit the conventional education mold due to cultural, language, or learning differences.

Educators benefit from SciQuiry as well. They will have access to detailed analytics for each student, which helps track progress and identify areas needing support. Our platform also provides educators with a rich library of questions that can be integrated into their teaching, allowing them to customize lessons to better fit the needs of their students.

By providing a platform that adapts to and supports the diverse needs of all students, SciQuiry aims to improve educational outcomes, foster a love for learning, and equip students with the skills necessary for academic and career success in STEM fields. This approach not only enhances individual learning experiences but also supports educators in their critical roles, making the educational process more effective and inclusive.

Rashin Taheri, the co-founder and CEO of SciQuiry, immigrated to the United States in 2001 as a woman in STEM. Her firsthand experience with the educational barriers faced by women in underrepresented communities has shaped her commitment to making STEM education accessible and inclusive. While volunteering at the Immigrant and Refugee Outreach Center (IROC) in the DMV area, Rashin witnessed the educational disparities that affect these groups. This insight drives her vision for SciQuiry: to enhance STEM education through technology and AI, with a particular emphasis on supporting girls and underrepresented students.

SciQuiry was founded in 2021 by Rashin and Dr. Hamid Taheri, combining their deep passion for education with extensive expertise in technology, healthcare, and entrepreneurship. Rashin brings more than two decades of experience in IT, application development, and management, having previously developed Koantum.com, an award-winning science platform for young learners. Dr. Hamid Taheri contributes his entrepreneurial skills and operational experience from establishing a successful cardiology practice and a clinical research program.

Our team’s vision is supported by guidance from esteemed advisors like Prof. Bruce McLaren from Carnegie Mellon University and Dr. Larry Medsker from George Washington University. Their expertise in artificial intelligence, educational technology, and the societal impacts of AI ensures that SciQuiry’s developments are not only cutting-edge but also ethically responsible.

SciQuiry has earned recognition and support, including a 2023 Department of Education SBIR Grant and microgrants from the VELA Education Fund and 4.0 Schools in 2022. 

We actively engage with a diverse range of students and educators, from high-performers to those who struggle, ensuring our platform is inclusive and responsive to their needs. 

SciQuiry is in the Pilot stage, having evolved from an MVP to extensive testing in educational settings. Insights from over 100 interviews under the Innovation Commercialization Assistance Program shaped the initial design, highlighting the need for better science exposure. Our pilot launched through the 4.0 Schools Fellowship, involved 200 middle school students and demonstrated that over 70% found the platform beneficial for understanding complex physics concepts. A Department of Education SBIR Grant in 2023 helped develop our SmartSeqQ algorithm, further tested with 115 students, showing significant efficacy. To date, SciQuiry has engaged over 315 students, significantly enhancing their STEM education.

Our platform is online and can be accessed from anywhere, making it available nationally and even globally. This ensures that students and educators from different regions can use it without any geographical limitations.

Recently, we conducted a pilot as part of the Department of Education grant with schools in the DMV area (Washington DC, Maryland, and Virginia). We are actively expanding our reach; for the 2024-25 school year, we have launched a cohort program that includes the Kingsway Regional School District in New Jersey and Snowline Joint Unified School District in California, along with 10 other charter schools across various states.

Unlike traditional educational models that treat all learners the same, SciQuiry adapts to the individual pace and learning style of each student, creating a personalized educational journey that keeps students engaged and improves their learning outcomes.

Our platform uses a distinctive question-based learning method, which research in educational psychology has shown to be more effective for retaining information than passive study methods. This approach is entirely driven by questions, setting it apart from other platforms in the market, which mix teaching content with assessments. By focusing solely on a question-driven curriculum, SciQuiry enhances STEM learning by applying theoretical knowledge in practical, real-world contexts.

At the heart of our innovation is the SmartSeqQ algorithm, an advanced version of the Deep Knowledge Tracing model, which personalizes the learning experience far more accurately than typical AI tools. It selects questions uniquely suited to each student’s learning progress, effectively simulating a personal tutor. This personalized approach is especially beneficial in under-resourced schools where educational resources are often limited.

SciQuiry also integrates generative AI through the ChatGPT API to dynamically create educational content, such as tailored questions and detailed explanations. This capability allows for continuous content renewal, ensuring that educational materials are always current and customized to meet students' needs without substantial manual effort.

Moreover, SciQuiry is designed with inclusivity at its core. It includes AI accessibility features such as text-to-speech and speech-to-text to support students with disabilities, like those with visual impairments or dyslexia. This makes STEM education more accessible to all students, demonstrating our commitment to creating a learning environment where every student has the opportunity to succeed.

SciQuiry leverages advanced Machine Learning (ML) and Deep Learning (DL) techniques to provide a personalized learning experience based on the principles of Knowledge Tracing (KT). 

Knowledge Tracing (KT) predicts future student interactions from historical performance data. Our implementation utilizes Deep Knowledge Tracing (DKT), an evolution from Bayesian Knowledge Tracing (BKT) that uses machine learning to more accurately model and predict student understanding. SciQuiry initiated the Smart Sequencing of Questions (SmartSeqQ) algorithm using DKT as a foundation. Essentially SmartSeqQ goes beyond predicting the probability of a student correctly answering questions in the database, as per DKT, to actually picking the next question from our database to present to the student. This AI-driven approach considers each student's unique response pattern to dynamically adjust the learning path, aiming for an optimal inquiry-based learning experience. Such a method stands to significantly enhance engagement and comprehension by presenting content that students are ready to learn, thereby maximizing the educational impact within the limited time available for learning.

To complement this, SciQuiry integrates Natural Language Processing (NLP) through the use of generative AI technologies, particularly the ChatGPT API into our platform. This integration allows for the automatic generation of customized educational content such as questions, explanations, and hints that are contextually relevant and tailored to each student's current understanding. All generated content undergoes a stringent review process by subject matter experts to ensure its accuracy and relevance.

Furthermore, SciQuiry incorporates advanced speech recognition and text-to-speech technologies to make learning accessible to all students, including those with disabilities like visual impairments or dyslexia. These technologies transform text into speech and vice versa, providing an interactive learning experience that accommodates various learning needs and preferences.

The technological infrastructure of SciQuiry is built for scalability and high performance, utilizing AWS cloud services to ensure reliability and security. The platform uses TypeScript for robust front-end development, AWS Cognito for secure user authentication, DynamoDB for efficient data management, and serverless GraphQL Lambdas for responsive backend operations. 

The effectiveness of SciQuiry's core technology, the Smart Sequencing of Questions (SmartSeqQ) algorithm, which leverages principles from Deep Knowledge Tracing (DKT), is supported by a combination of our internal data, pilot study outcomes, and corroborating academic research. 

During Phase I of the Department of Education (DOE) grant, SciQuiry partnered with Carnegie Mellon University to utilize their implementation of Bayesian Knowledge Tracing (BKT) as a foundational proof of concept. This partnership allowed us to refine and develop our SmartSeqQ algorithm, an advanced implementation of Deep Knowledge Tracing (DKT). These algorithms are specifically designed to model and adapt to individual student understanding, intelligently selecting questions to address students' specific learning needs and misconceptions.

Pilot Study Outcomes:
Our recent Phase I pilot study involved 115 middle school students and aimed to assess the effectiveness of the SciQuiry algorithm compared to the adaptive algorithms from Carnegie Mellon University. Descriptive analyses indicated that students using the SciQuiry platform showed improvements in their science performance. There was also a noticeable decrease in error rates among students, suggesting that the platform effectively enhanced their understanding and retention of material.

Academic Research and Validation:
Knowledge Tracing (KT) principles guide the prediction of future student interactions based on historical performance data (Corbett & Anderson, 1995), supporting personalized learning paths. This customization is essential, as research by Bloom (1968) indicates significant learning gains when instruction is tailored to the learner's pace and knowledge level (Vygotsky, 1978).

Deep Knowledge Tracing (DKT), a more recent algorithm and advancement from the original Bayesian Knowledge Tracing (BKT) algorithm, employs machine learning to model and predict student understanding more accurately (Piech et al., 2015). Unlike BKT, DKT's ability to forecast a student's knowledge state for future interactions enables more precise customization of content and difficulty level, ensuring that each student engages with material that is optimally challenging. SciQuiry initiated the Smart Sequencing of Questions (SmartSeqQ) algorithm using DKT as a foundation. Essentially SmartSeqQ goes beyond predicting the probability of a student correctly answering questions in the database, as per DKT, to actually picking the next question to present to the student. This approach can thus maximize the student’s learning efficiency. 
This AI-driven approach considers each student's unique response pattern to dynamically adjust the learning path, aiming for an optimal inquiry-based learning experience. Such a method stands to significantly enhance engagement and comprehension by presenting content that students are ready to learn, thereby maximizing the educational impact within the limited time available for learning.

As the co-founder of SciQuiry and a volunteer with the Immigrant and Refugee Outreach Center (IROC) in the DMV area, I have firsthand experience with the educational barriers faced by underrepresented communities. This experience has been pivotal in shaping the development of SciQuiry. We actively incorporate feedback from these communities into our platform’s design and content. Our upcoming cohort for the 2024-25 school year will include a diverse group of educators and students who will participate in shaping future developments, ensuring our solutions are tailored to the needs of those most in need of support.

Our platform benefits significantly from the insights of esteemed advisors specializing in AI ethics and educational technology. Professor Bruce McLaren from Carnegie Mellon University and Dr. Larry Medsker from George Washington University guide our AI implementations. Their expertise ensures that our developments are not only cutting-edge but also ethically grounded, focusing on the societal impacts of AI. Their contributions are critical in helping us design algorithms that are fair and provide equitable learning opportunities.

To ensure our content development process is free from bias, our developers come from varied backgrounds and are trained to recognize and eliminate bias in educational materials. We train our AI systems using a wide range of data that reflect the diversity of our user base, helping to prevent the reinforcement of existing biases. All educational content and algorithmic decisions are regularly reviewed by a panel that includes our AI advisors and community representatives. This continuous oversight helps identify and correct any biases that may arise.

Full-time: 3

• Rashin Taheri, Co-founder and CEO
• Kyle Reto, AI & Machine Learning Engineer
• Olzhas Alexandrov, Full Stack Developer

Part-time: 2

• Dr. Hamid Taheri, Co-founder and CFO
• Dr. Anne Spear, Business Developer, focusing on product commercialization and impact

Consultants: 2

• Prof. Bruce M. McLaren, Researcher and Advisor
• Dr. Karen Medsker, Learning Expert and Instructional Designer

In 2023, SciQuiry secured a highly competitive $250K grant from the Department of Education's SBIR Phase I program, which initiated the development of our SmartSeqQ algorithm. As part of this grant, we conducted a pilot study in February 2024 involving 115 students from Basis Independent McLean and DC International School. This study assessed the effectiveness of our proprietary SmartSeqQ algorithm against alternative adaptive algorithms, providing valuable insights into enhancing our AI-driven personalization techniques.

Building on this momentum, we are now focusing on developing a teacher dashboard, enriching our platform with a variety of question types and an expanded general science content for grades 3-8. By integrating these new features, SciQuiry is gearing up to be pilot-ready for the 2024-25 school year.

Our advancements this year position us well to start pilot tests in classrooms with priority learners, particularly focusing on Transitional Kindergarten to Grade 8 students, including those from Black and Latino backgrounds and students experiencing poverty. 

SciQuiry operates on a freemium business model, which allows us to provide significant educational value to a broad range of users while also generating the revenue necessary to sustain and expand our services. 

Revenue will be generated from a one-time licensing fee for schools and districts, and a subscription model for parents seeking premium features. 

The licensing fee for educational institutions is set at $8 per student per year, providing full access to the core curriculum and basic platform functionalities. This fee is deliberately set low to keep the platform affordable for public schools, particularly those in under-resourced areas.

For parents desiring an enhanced learning experience for their children, a monthly subscription of $7 per user offers additional premium features. These features include customized content delivery through adaptive learning pathways, detailed progress reports, and AI-driven recommendations for expanded learning at home.

Educators receive free access to our comprehensive dashboard, which allows them to monitor student progress in real-time and adjust instruction based on detailed analytics. This support extends to the integration of our platform into school curriculums, facilitating a seamless blend of in-class and remote learning. Schools purchasing subscriptions for groups of students receive bulk discounts, making it more affordable to provide enhanced learning tools to every student.

To ensure scalability, SciQuiry is built on a robust cloud infrastructure that supports rapid expansion and reliable access from anywhere. This technical backbone allows us to maintain high performance as usage grows, ensuring that our platform remains accessible to a large number of users simultaneously without degradation in service quality.

We are applying to the Learner//Meets//Future Challenge because it aligns perfectly with SciQuiry's mission to enhance student learning through innovative, AI-driven approaches. This challenge provides us with essential resources and mentorship to help expand our technology and increase our impact.

Refining our SmartSeqQ algorithm to better personalize learning experiences is a key technical priority. Collaborating with Solve’s network of technology experts could speed up this process, allowing us to add more advanced features and ensure our AI performs well in diverse learning environments.

As our educational technology platform grows, we face challenges in broadening our user base and entering new markets. We are particularly focused on reaching diverse educational settings, including schools in low-income and underserved areas where our technology could have a significant impact. Through Solve, we hope to connect with partners and experts experienced in the ed-tech market, who can offer strategic guidance on scaling and distributing our platform effectively, especially in under-resourced areas.